Magnus only matched the top engine move 43% of the time, which is quite low for a super-GM. Hikaru said he'd bet money that it was Magnus' worst performance in any game in the last 3 years.
If the evidence shows that Magnus played badly, we do not have to jump to the conclusion that actually he was playing badly on purpose, for the intent of trying to detect a cheater. He himself wasn't even stupid enough to claim this.
No, instead, I am going to trust other grandmasters opinions that he was merely playing badly.
(Sorry for the dupe reply, too late to edit my other comment)
I agree but Magnus also plays bad games once in a while. So it is not quite clear from a sample size 1 and a std deviation of infinite.
There are a number of comments here which promote the idea that this is simply sour grapes by Magnus and totally uncalled for, and I find those perspectives to be overwhelmingly uncompelling.
Of course he played near-perfectly for 27 moves. That's what it means to be a 2700 (or even a 2500, if you think his recent ratings growth is the product of cheating). Virtually any high-level game will look like that - the players playing mostly perfectly, with on average one or two significant mistakes, but sometimes not even that.
To be clear, I'm not saying this to make a claim that this is definitive proof of anything. I'm pointing out that the theory that Magnus simply blundered away the game doesn't hold water. Niemann had the advantage—as black no less—for essentially the entire game against someone who is widely known for being capable of grinding away nearly-perfectly for extensive periods of time.